rabi3elbeji / jddposter2018

Poster published in the doctoral day at ENIT 2018

Journée des doctorants (JdD ENIT 2018)

Model architecture

Experiment Results

• This repository contains an implementation of image steganalysis based on the convolutional neural network.

• The accuracy of model classification cross 95 % in a high payload (for 1bpp).

• We use WOW as a steganographic algorithm to evaluate our model.

• In the figure below the training process on a high payload: 1bpp.

• For the payload less than 1 bpp, we used the transfer learning to accelerate the model convergence.

• Experiment results of WOW :

| payload (bpp) | 1.0 | 0.9 | 0.8 | 0.7 | 0.6 | 0.5 | | --- | --- | --- | --- | --- | --- | --- | --- | | Precision | 95.12% | 94.25% | 92.50% | 89.75% | 87.36% | 83.25% |

Requirement Software:

• Tebsorflow v > 1.0
• Keras v > 2.0

Authors:

• Rabii Elbeji

License

MIT